Great explanation, btw.
Great explanation, btw.
http://hebb.mit.edu/courses/9.29/2003/athena/bearlee/Neural_...
An integrator basically solves the ODE y'(t)=f(t); the firing rate is proportional to y(t). So if you wire up integrators appropriately, you could solve most ODE's (obviously, there are scaling issues, noise, etc). In principle, one could reproduce the bee's behavior by appropriately connecting six of these.
But I admit, this example is one of the more fantastic examples I've seen. It's a simple algebraic manifold (1) and obtained purely from observation of behavior rather than lab work.
(1) In biology, you usually get shapes which are big ugly messes, imagine the surface of a child's playdough sculpture. Instead, she found something pretty like kepler's conic sections (to borrow the analogy from the article). That's uncommon in biology.